Diesel generating set shutdown detection and control system

Abstract

The invention discloses a shut-down detection and control system for a diesel generating set, belongs to the technical field of monitoring, and solves the problem that the service life of the diesel generating set is affected by damage to internal results caused by direct shut-down of the diesel generating set due to the fact that clamping stagnation of a piston ring in the diesel generating set cannot be detected through a sensor. Comprising a global data acquisition and space-time synchronization module, a data processing module and a communication module, the data processing module receives multi-sensor data flow from the global data acquisition and time-space synchronization module after synchronous clock alignment and diagnoses the clamping stagnation phenomenon of the piston ring through a full-dimensional dynamic fusion model; and the communication module is used for outputting the clamping stagnation phenomenon of the piston ring diagnosed by the data processing module. According to the method, high-precision diagnosis of the clamping stagnation fault of the piston ring of the diesel generating set can be realized through multi-source heterogeneous data fusion, nonlinear dynamic modeling and cross-domain coupling analysis.

Description

Technical Field

[0001] The invention relates to the technical field of monitoring, and in particular to a shutdown detection and control system for a diesel generator set. Background Art

[0002] Diesel generator set is a device that uses diesel engine as power to drive generator to generate electricity. It is widely used in industrial, commercial, construction sites, emergency rescue and remote areas. It consists of diesel engine, generator, control system and supporting equipment, and has the characteristics of high efficiency, stability and reliability. As a power source, diesel engine has strong power output and high fuel economy, and can operate stably in various environments. The generator converts mechanical energy into electrical energy to meet different power requirements. The control system can realize functions such as automatic start, shutdown and protection to ensure the safe operation of the equipment. In addition, diesel generator set also has the advantages of strong mobility, easy installation and low maintenance cost. Especially in the case of unstable power grid or no power supply, it can quickly provide power support to ensure the operation of equipment and production and life electricity, and is a reliable backup power supply option.

[0003] Before shutting down a diesel generator set, it is necessary to test various indicators in the diesel generator set to ensure that its shutdown indicators meet the standards. Otherwise, the unit will be damaged after it is shut down directly because all indicators of the diesel generator set are not lower than the threshold.

[0004] Among them, most indicators can feedback direct data through sensor detection, and the accuracy of the data can be determined through environmental compensation algorithms. However, there are also problems in diesel generator sets that cannot be detected by sensors, such as piston ring sticking.

[0005] If the diesel generator set has piston ring sticking before shutdown, it is mainly due to combustion carbon deposits, oil carbonization or impurity intrusion, which causes the piston ring to lose elasticity. During operation, the cylinder sealing will decrease, causing gas leakage and blowby to the crankcase, resulting in oil dilution and abnormal pressure. At the same time, excessive oil will enter the combustion chamber, causing blue smoke from the exhaust pipe and a sharp increase in oil consumption. The heat from the piston head cannot be effectively transferred to the cylinder wall. Local overheating may cause piston burning or cylinder wall strain, which manifests as power reduction, difficulty in starting and abnormal noise during operation. If the unit is shut down directly without timely processing, the residual high temperature will accelerate the hardening of carbon deposits and further solidify the stuck state. Long-term accumulation may cause piston ring breakage or serious cylinder strain, and ultimately the unit needs to be disassembled and overhauled to restore performance.

[0006] Therefore, it is of certain significance if the piston ring sticking can be detected before the diesel generator set is shut down, and then whether the diesel generator set can be shut down to avoid damage to the set.

[0007] Therefore, a diesel generator set shutdown detection and control system is proposed to solve or alleviate the above problems. Summary of the invention

[0008] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a diesel generator set shutdown detection and control system.

[0009] In order to achieve the above object, the present invention adopts the following technical solutions: A diesel generator set shutdown inspection and control system includes a global data acquisition and spatiotemporal synchronization module, a data processing module, and a communication module. The global data acquisition and spatiotemporal synchronization module defines a global synchronous clock-aligned multi-sensor data stream after data acquisition through sensors. The data processing module receives the multi-sensor data stream from the global data acquisition and the spatiotemporal synchronization module after synchronization clock alignment and diagnoses the piston ring sticking phenomenon through a full-dimensional dynamic fusion model. The communication module is used to output the piston ring sticking phenomenon diagnosed by the data processing module.

[0010] Preferably, the global data acquisition and spatiotemporal synchronization module defines a global synchronization clock to align multi-sensor data streams after performing data acquisition through sensors, comprising the following steps: Synchronous acquisition of mechanical vibration signals of diesel generator sets , cylinder wall temperature distribution , Crankcase blowby flow , exhaust multi-component spectrum , and crankshaft angle-cylinder pressure phase delay signal , the mechanical vibration signal is the high-frequency vibration acceleration of the cylinder body, the sampling frequency of the high-frequency vibration acceleration of the cylinder body is ≥50kHz, the spatial resolution is ≤0.1g, the cylinder wall temperature distribution is collected by 16-point gridding, and the temperature resolution is ≤0.1℃; Define a global synchronization clock , aligning multi-sensor data streams through the PTP protocol; Establish the cylinder wall temperature field grid coordinate system , correlated vibration signal phase .

[0011] Preferably, the data processing module receives the multi-sensor data stream from the global data acquisition and the time-space synchronization module after synchronization clock alignment, and diagnoses the piston ring sticking phenomenon through the full-dimensional dynamic fusion model, including the following steps: The mechanical vibration signal is decomposed by wavelet packets to extract the energy of high-frequency sensitive frequency bands, and the cylinder wall temperature field gradient tensor and hot spot diffusion index are calculated; Construct nonlinear state-space equations to describe the interaction between mechanical vibration, thermodynamic fields and combustion chemistry, and identify cross-domain coupling coefficients online; Calculate the mutual information entropy of vibration energy and temperature gradient, and construct the dynamic transfer function of blowby flow rate and unburned hydrocarbon concentration; Based on primary suspicious index screening, intermediate spatiotemporal delay verification and advanced tensor decomposition, the diagnosis result of piston ring sticking is output; The coupling coefficient is updated online through a recurrent neural network to optimize the cost-sensitive loss function of the fault confusion matrix.

[0012] Preferably, the step of performing wavelet packet decomposition on the mechanical vibration signal to extract high-frequency sensitive band energy and calculating the cylinder wall temperature field gradient tensor and hot spot diffusion index comprises the following steps: The mechanical vibration signal is decomposed by 8-layer wavelet packet to extract the energy characteristics of the piston ring sticking sensitive frequency band: ,in, For the wavelet system of the mth layer and the nth node, the sensitive frequency band corresponds to m=7, n=32-47, and the piston ring sticking sensitive frequency band is 2.5-4.2kH; Calculate the Frobenius norm of the cylinder wall temperature gradient tensor and generate the hot spot diffusion index ,in, is the heat conduction attenuation coefficient, and N is the number of effective grids.

[0013] Preferably, the nonlinear state space equation includes a mechanical vibration subsystem and a thermodynamic subsystem, wherein the mechanical vibration subsystem includes ,in, is the mechanical vibration state variable, is the coupling term of thermal field and chemical field to mechanical vibration. The thermodynamic subsystem includes ,in, is the thermal diffusion coefficient, is the influence coefficient of blowby gas flow rate on temperature, is the influence coefficient of vibration energy on temperature. The cross-domain coupling coefficient identification uses Kalman filtering to estimate the coupling matrix online. ,in, and Represent the state variables of thermal field and chemical field respectively.

[0014] Preferably, the calculation of the mutual information entropy of the vibration energy and the temperature gradient and the construction of the dynamic transfer function of the blowby gas flow rate and the unburned hydrocarbon concentration comprises the following steps: Calculate the mutual information entropy of the vibration energy and the temperature gradient, the mutual information entropy of the temperature gradient is the mutual information entropy of the hot spot diffusion index, ,like , it indicates that the mechanical-thermal coupling is abnormal; Constructing the dynamic transfer function of blowby gas flow rate and unburned HC concentration , a significant resonance peak appears in the 0.5-2Hz frequency band , indicating that there is a combustion abnormality.

[0015] Preferably, the method of outputting the diagnosis result of piston ring sticking based on primary suspicious index screening, intermediate spatiotemporal delay verification and advanced tensor decomposition comprises the following steps: Primary screening calculation of piston ring sticking suspicion index ,in, is the weight coefficient, and is the reference value under normal working conditions. , then it is suspected that there is a piston ring sticking fault; Intermediate verification of the propagation delay of vibration energy and hot spot diffusion index ,like , eliminating transient interference and confirming it is a continuous fault; Advanced diagnostics builds a 4D fault feature tensor , using Tucker decomposition to extract core failure modes , Fault contribution ,like And other modes , it is diagnosed as a piston ring sticking fault.

[0016] Preferably, the online updating of the coupling coefficient by a recursive neural network to optimize the cost-sensitive loss function of the fault confusion matrix comprises the following steps: Updating coupling coefficients using recurrent neural networks ,in, and It is the state change of mechanical vibration and thermal field; Defining cost-sensitive loss function , give priority to reducing the false negative rate and complete the fault confusion matrix optimization.

[0017] The present invention has the following beneficial effects: The present invention realizes high-precision diagnosis of piston ring sticking and complex fault differentiation through multi-source heterogeneous data fusion, nonlinear dynamic system modeling and cross-domain covariance topological analysis. Global data acquisition and time-space synchronization ensure the precise alignment of multi-dimensional data, providing a reliable basis for subsequent analysis. Signal preprocessing and feature enhancement can effectively extract sensitive features of piston ring sticking, such as specific frequency band energy in vibration signals and hot spot diffusion index of temperature fields, which are highly sensitive to faults. The multi-physics field dynamic coupling modeling reveals the intrinsic connection between mechanical vibration, thermodynamics, fluid mechanics, combustion chemistry and other multi-physics fields. Through the nonlinear state space equation and cross-domain coupling coefficient identification, the propagation law and interaction mechanism of piston ring sticking fault in multi-physics fields are accurately described. The deep extraction of fault features further explores the vibration-thermal field joint entropy characteristics and the blowby-combustion chemistry correlation map. These features can reflect the essential characteristics of piston ring sticking fault from different angles. Multi-level decision fusion integrates the results of primary feature screening, intermediate spatiotemporal correlation verification and advanced tensor decomposition, and finally achieves accurate diagnosis of piston ring sticking fault through multi-level judgment basis such as piston ring sticking suspicion index, vibration-thermal field propagation delay analysis and fault contribution. It can adapt to the operation changes of diesel generator sets under different working conditions, and continuously improve the accuracy of diagnosis through online parameter self-correction and fault confusion matrix optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 It is a structural block diagram of the present invention.

[0020] 1. Global data acquisition and spatiotemporal synchronization module; 2. Data processing module; 3. Communication module. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0022] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0024] In the description of the present invention, it should be understood that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, or are the orientations or positional relationships in which the product of the invention is conventionally placed when in use, or are the orientations or positional relationships conventionally understood by those skilled in the art. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0025] Furthermore, the terms “first”, “second”, “third”, etc. are merely used for distinguishing descriptions and are not to be understood as indicating or implying relative importance.

[0026] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0027] A diesel generator set shutdown detection and control system, such as Figure 1 As shown, it includes a global data acquisition and spatiotemporal synchronization module 1, a data processing module 2, and a communication module 3. The global data acquisition and spatiotemporal synchronization module 1 defines a global synchronous clock-aligned multi-sensor data stream after data acquisition through sensors. The data processing module 2 receives the multi-sensor data stream from the global data acquisition and spatiotemporal synchronization module after synchronization clock alignment and diagnoses the piston ring sticking phenomenon through a full-dimensional dynamic fusion model. The communication module 3 is used to output the piston ring sticking phenomenon diagnosed by the data processing module 2.

[0028] This system uses the sensor group in the global data acquisition and spatiotemporal synchronization module 1, which includes accelerometers, temperature sensors, and spectrometers to collect key parameters such as mechanical vibration signals, cylinder wall temperature distribution, and crankcase blowby flow, to ensure the comprehensiveness and real-time nature of the data. It uses PTP to define the global synchronization clock, eliminates the time deviation of multi-sensor data streams, and provides an accurate spatiotemporal alignment basis for subsequent analysis. The data processing module 2 uses a full-dimensional dynamic fusion model, combined with wavelet packet decomposition, nonlinear state space modeling, mutual information entropy calculation, and tensor decomposition technologies, to extract fault features from multiple dimensions, and update model parameters online through recursive neural networks. It dynamically optimizes fault diagnosis logic to adapt to changes under different working conditions, and outputs the diagnosis results to the control terminal or cloud platform in real time to support operators in making quick decisions. Collaborative analysis of multi-physics field data can significantly improve the comprehensiveness and accuracy of fault detection through joint analysis of mechanical vibration, thermal field, fluid, and chemical data. The vibration signal can capture the friction characteristics of piston ring sticking, and the temperature field gradient can reflect local thermal stress anomalies. The combination of the two can avoid misjudgment. The high-frequency dynamic modeling capability of mechanical vibration signals is sampled at ≥50kHz, and the temperature field is collected in a 16-point grid, which can capture the transient characteristics of piston ring sticking and achieve early warning. The PTP protocol is used to achieve microsecond synchronization of multi-sensor data, which can ensure the time consistency of data even under variable load or extreme temperature environments, avoiding diagnostic errors caused by timing misalignment in traditional methods. In addition, the system is combined with an online learning mechanism, which can dynamically adjust the diagnostic thresholds and weights according to real-time data to solve the problem of insufficient adaptability of traditional fixed threshold models under complex working conditions.

[0029] Preferably, the global data acquisition and spatiotemporal synchronization module 1 defines a global synchronization clock to align multi-sensor data streams after acquiring data through sensors, including the following steps: Synchronous acquisition of mechanical vibration signals of diesel generator sets , cylinder wall temperature distribution , Crankcase blowby flow , exhaust multi-component spectrum , and crankshaft angle-cylinder pressure phase delay signal ,The mechanical vibration signal is the high frequency vibration acceleration of the cylinder body, the sampling frequency of the high frequency vibration acceleration of the cylinder body is ≥50kHz, the spatial resolution is ≤0.1g, and the cylinder wall temperature distribution is collected through 16-point gridding, and the temperature resolution is ≤0.1℃; Define a global synchronization clock , aligning multi-sensor data streams through the PTP protocol; Establish the cylinder wall temperature field grid coordinate system , correlated vibration signal phase .

[0030] Through the above method steps, the 50kHz sampling rate can analyze the micro-vibration caused by piston ring sticking. Traditional methods are prone to miss such high-frequency features. The 0.1g resolution can detect changes in tiny friction forces to avoid masking the real fault characteristics due to signal noise. The 16-point grid covers the key areas of the cylinder wall, specifically the piston ring movement path. The 0.1℃ resolution can identify local overheating and provide a data basis for thermal stress analysis. The PTP protocol controls the time deviation of multiple sensors within 100μs to ensure strict synchronization of vibration signals and temperature field changes. The vibration peak caused by piston ring sticking can be accurately correlated with the sudden change in temperature gradient, avoiding misjudgment caused by timing misalignment in traditional methods.

[0031] Preferably, the data processing module 2 receives the multi-sensor data stream from the global data collection and the time-space synchronization module after synchronization clock alignment, and diagnoses the piston ring sticking phenomenon through the full-dimensional dynamic fusion model, including the following steps: The mechanical vibration signal is decomposed by wavelet packets to extract the energy of high-frequency sensitive frequency bands, and the cylinder wall temperature field gradient tensor and hot spot diffusion index are calculated; Construct nonlinear state-space equations to describe the interaction between mechanical vibration, thermodynamic fields and combustion chemistry, and identify cross-domain coupling coefficients online; Calculate the mutual information entropy of vibration energy and temperature gradient, and construct the dynamic transfer function of blowby flow rate and unburned hydrocarbon concentration; Based on primary suspicious index screening, intermediate spatiotemporal delay verification and advanced tensor decomposition, the diagnosis result of piston ring sticking is output; The coupling coefficient is updated online through a recurrent neural network to optimize the cost-sensitive loss function of the fault confusion matrix.

[0032] The above method steps subdivide the vibration signal into a sensitive frequency band of 2.5-4.2kHz through an 8-layer decomposition, effectively filter out background noise, specifically the low-frequency vibration of other engine components, improve the signal-to-noise ratio, and have the frequency domain advantage of wavelet packet decomposition. Through the adaptability of the nonlinear model, the state-space equation associates mechanical vibration and thermodynamic field through coupling terms, and can dynamically reflect the changes in the energy transfer path caused by piston ring sticking. In addition, through the joint analysis capability of mutual information entropy, if the piston ring sticking causes the sealing to decrease, the mutual information entropy of vibration energy and hot spot diffusion index will increase significantly, avoiding single parameter misjudgment, and the composite fault decoupling Tucker decomposition of tensor decomposition decomposes the four-dimensional fault feature tensor, which includes time, frequency, space, and chemistry, into core patterns, accurately distinguishing piston ring sticking from injector blockage faults.

[0033] Preferably, the mechanical vibration signal is subjected to wavelet packet decomposition to extract the high-frequency sensitive band energy, and the cylinder wall temperature field gradient tensor and hot spot diffusion index are calculated, including the following steps: The mechanical vibration signal is decomposed by 8-layer wavelet packet to extract the energy characteristics of the piston ring sticking sensitive frequency band: ,in, For the wavelet system of the mth layer and nth node, the sensitive band corresponds to m=7, n=32-47, is the vibration energy extracted in the piston ring sticking sensitive frequency band after the mth layer wavelet packet decomposition, is the wavelet coefficient of the nth node in the mth layer, indicating the intensity of the signal at a specific frequency f and time t. and is the high frequency upper limit and low frequency lower limit of the sensitive frequency band in the mth layer of wavelet packet decomposition. The sensitive frequency band of piston ring sticking is 2.5-4.2 kHz. n is the index of the node in the wavelet packet decomposition, which is used to traverse all possible frequency sub-bands. Calculate the Frobenius norm of the cylinder wall temperature gradient tensor and generate the hot spot diffusion index ,in, is the heat conduction attenuation coefficient, N is the effective grid number, For the spatial position and the cylinder wall temperature gradient tensor constructed at time t, is the hot spot diffusion index, which is used to quantify the diffusion trend of cylinder wall temperature change in space. N is the number of effective grids, which is the number of cylinder wall temperature grid points involved in the calculation. is the trace of the temperature gradient tensor, that is, the sum of the diagonal elements, which is used to comprehensively reflect the overall trend of temperature change.

[0034] Through the above method steps, the 2.5-4.2kHz frequency band is the characteristic frequency range of piston ring sticking by focusing the energy of the sensitive frequency band. The traditional FFT analysis is easily interfered by broadband noise, while the wavelet packet decomposition can accurately extract the energy of this frequency band and improve the fault detection sensitivity. The first-order and second-order temperature gradient information are integrated through the Frobenius norm, which can more comprehensively reflect the abnormal thermal stress distribution and avoid the limitation of single-dimensional analysis. The exponential attenuation term in the index can suppress the interference of the steady-state thermal field and highlight the transient hot spot, which is specifically the diffusion characteristic of the periodic friction heat caused by piston ring sticking.

[0035] Preferably, the nonlinear state space equation includes a mechanical vibration subsystem and a thermodynamic subsystem, and the mechanical vibration subsystem includes ,in, is the mechanical vibration state variable, is the coupling term of thermal field and chemical field to mechanical vibration. The thermodynamic subsystem includes ,in, is the thermal diffusion coefficient, is the rate of change of cylinder wall temperature with time, is the Laplace operator of temperature, which indicates the diffusion of temperature in space. is the crankcase blowby flow rate, is the rate of change of piston ring vibration energy over time, is the influence coefficient of blowby gas flow rate on temperature, The cross-domain coupling coefficient is identified by using Kalman filtering to estimate the coupling matrix online. ,in, To find the coupling matrix that minimizes the objective function, is the objective function, which represents the sum of the squares of the error between the state change rate of the mechanical vibration subsystem and the coupling matrix acting on the state vectors of the thermodynamic and chemical subsystems, and is used to optimize the coupling matrix. and denote the state variables of thermal and chemical fields respectively, is the rate of change of the state vector of the mechanical vibration subsystem over time, is the state matrix of the mechanical vibration subsystem, describing the dynamic relationship between the internal states of the system. is the state vector of the mechanical vibration subsystem and contains Piston ring vibration energy, Vibration signal phase, Vibration acceleration standard deviation, is the input matrix of the mechanical vibration subsystem, describing the influence of the input on the system state. is the input vector of the mechanical vibration subsystem, which may include external excitation, etc. is the thermal-chemical coupling term, which represents the coupling effect of thermodynamic field and chemical reaction on the mechanical vibration subsystem.

[0036] Through the above method steps, accurate modeling of multi-field coupling is performed. The traditional single-field model cannot describe the multi-physical effects of piston ring sticking, and this equation quantifies the thermal-vibration-chemical interaction through coupling terms, specifically the immediate impact of friction heat on the temperature field, thereby improving the prediction accuracy of the model. At the same time, the above method steps have the robustness of online parameter adaptation. The Kalman filter is dynamically updated according to real-time data to adapt to parameter changes under different loads or ambient temperatures, such as changes in thermal diffusion coefficient under high temperature, thereby avoiding the limitations of the offline calibration model. The acceleration standard deviation can reflect the randomness of the piston motion, such as the non-periodic impact caused by sticking, and together with the energy and phase, constitute a complete description of the vibration state.

[0037] Preferably, calculating the mutual information entropy of the vibration energy and the temperature gradient and constructing a dynamic transfer function of the blowby gas flow rate and the unburned hydrocarbon concentration comprises the following steps: Calculate the mutual information entropy of vibration energy and temperature gradient. The mutual information entropy of temperature gradient is the mutual information entropy of hot spot diffusion index. ,like , indicating that the mechanical-thermal coupling is abnormal, is the mutual information entropy between the vibration energy and the hot spot diffusion index, is the joint probability density function of the vibration energy E and the hot spot diffusion index K, is the marginal probability density function of the vibration energy E, is the marginal probability density function of the hot spot diffusion index K, is the threshold of mutual information entropy, which is used to determine whether the mechanical-thermal coupling is abnormal; Constructing the dynamic transfer function of blowby gas flow rate and unburned HC concentration , a significant resonance peak appears in the 0.5-2Hz frequency band , indicating that there is a combustion abnormality, is the Fourier transform of the blowby flow signal, is the Fourier transform of the unburned HC concentration signal, is the magnitude of the transfer function, It is three times the baseline standard deviation and is used to determine whether the transfer function amplitude is significantly abnormal.

[0038] In the above method steps, through the joint diagnosis advantage of mutual information entropy, when the piston ring is stuck and the sealing performance is reduced, the increase in vibration energy and the accelerated diffusion of hot spots have a time correlation. Mutual information entropy can quantify this nonlinear relationship to avoid the misjudgment of the single threshold method. The resonance peaks of the blowby flow rate and the unburned HC concentration in the 0.5-2Hz frequency band are typical characteristics of piston ring sticking due to periodic gas leakage, while fuel problems usually manifest as full-band abnormalities. The two can be effectively distinguished. Mutual information entropy is robust to noise and random fluctuations. The dynamic transfer function eliminates steady-state interference by focusing in the frequency domain. The joint use can significantly reduce the false alarm rate, thereby improving the anti-interference ability of the system.

[0039] Preferably, based on primary suspicious index screening, intermediate spatiotemporal delay verification and advanced tensor decomposition, outputting the piston ring sticking diagnosis result comprises the following steps: Primary screening calculation of piston ring sticking suspicion index ,in, is the weight coefficient, is the vibration energy of the piston ring, and is the reference value under normal working conditions. is the hot spot diffusion index, is the Frobenius norm of the coupling matrix, which is used to quantify the degree of coupling. , then it is suspected that there is a piston ring sticking fault; Intermediate verification of the propagation delay of vibration energy and hot spot diffusion index ,like , excluding transient interference, confirming that it is a continuous fault, among which, To find the delay time that makes the objective function reach the maximum value , is the vibration energy and hot spot diffusion index in the delay The correlation function below is used to measure the synchronization between the two; Advanced diagnostics builds a 4D fault feature tensor , using Tucker decomposition to extract core failure modes , fault contribution ,like And other modes , it is diagnosed as a piston ring jam fault, among which, is the space to which the tensor belongs, indicating the sizes of the tensor in the four dimensions of time, frequency, space and chemistry respectively. , , , , is the core tensor obtained by Tucker decomposition, which contains the main structural information of the fault characteristics. , , , The factor matrices corresponding to the time, frequency, space and chemical dimensions are used to decompose the original tensor into the feature space of each dimension. , , , Represents the product operation of tensor and matrix in corresponding dimensions, is the contribution of piston ring sticking fault, indicating the proportion of this fault mode in the overall fault characteristics. is the norm of the portion of the core tensor related to piston ring sticking failure, is the sum of the norms of all failure mode parts in the core tensor.

[0040] In the above method steps, the reliability of multi-level progressive diagnosis, the primary screening quickly narrows the scope of the fault, the intermediate verification eliminates transient interference through time series correlation, such as short-term load mutation, and the advanced diagnosis uses tensor decomposition to ensure the uniqueness of the results and avoid the risk of misjudgment in single-level decision-making. At the same time, the composite faults are decoupled through tensor decomposition. Since traditional PCA or ICA methods are difficult to process high-dimensional data, Tucker decomposition can separate fault modes in time, frequency, space, and chemical dimensions, and accurately distinguish between piston ring sticking and concurrent turbocharger failure.

[0041] Preferably, the coupling coefficient is updated online by a recursive neural network to optimize the cost-sensitive loss function of the fault confusion matrix, including the following steps: Updating coupling coefficients using recurrent neural networks ,in, and is the state change of mechanical vibration and thermal field, is the updated coupling matrix at the k+1th moment, is a recurrent neural network that is used to dynamically update the coupling matrix based on historical data and current changes. is the coupling matrix at the kth moment; Defining cost-sensitive loss function , give priority to reducing the false negative rate and complete the fault confusion matrix optimization, where, is the total cost of underreporting, is the total cost of false positives, is the cost of delayed diagnosis.

[0042] Through the above method steps, The online learning mechanism keeps the algorithm stable under harsh working conditions.

[0043] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A diesel generator set shutdown detection and control system, characterized in that: The invention comprises a global data acquisition and spatiotemporal synchronization module (1), a data processing module (2), and a communication module (3). The global data acquisition and spatiotemporal synchronization module (1) defines a multi-sensor data stream with global synchronous clock alignment after data acquisition through sensors. The data processing module (2) receives the multi-sensor data stream from the global data acquisition and spatiotemporal synchronization module after synchronous clock alignment and diagnoses the piston ring sticking phenomenon through a full-dimensional dynamic fusion model. The communication module (3) is used to output the piston ring sticking phenomenon diagnosed by the data processing module (2).

2. A diesel generator set shutdown detection and control system according to claim 1, characterized in that: The global data acquisition and spatiotemporal synchronization module (1) defines a global synchronization clock to align multi-sensor data streams after collecting data through sensors, including the following steps: Synchronous acquisition of mechanical vibration signals of diesel generator sets , cylinder wall temperature distribution , Crankcase blowby flow , exhaust multi-component spectrum , and crank angle-cylinder pressure phase delay signal , the mechanical vibration signal is the high-frequency vibration acceleration of the cylinder body, the sampling frequency of the high-frequency vibration acceleration of the cylinder body is ≥50kHz, the spatial resolution is ≤0.1g, the cylinder wall temperature distribution is collected by 16-point gridding, and the temperature resolution is ≤0.1℃; Define a global synchronization clock , aligning multi-sensor data streams through the PTP protocol; Establish the cylinder wall temperature field grid coordinate system , correlated vibration signal phase .

3. A diesel generator set shutdown detection and control system according to claim 2, characterized in that: The data processing module (2) receives the multi-sensor data stream from the global data collection and the time-space synchronization module after synchronizing the clocks and diagnoses the piston ring sticking phenomenon through the full-dimensional dynamic fusion model, including the following steps: The mechanical vibration signal is decomposed by wavelet packets to extract the energy of high-frequency sensitive frequency bands, and the cylinder wall temperature field gradient tensor and hot spot diffusion index are calculated; Construct nonlinear state-space equations to describe the interaction between mechanical vibration, thermodynamic fields and combustion chemistry, and identify cross-domain coupling coefficients online; Calculate the mutual information entropy of vibration energy and temperature gradient, and construct the dynamic transfer function of blowby flow rate and unburned hydrocarbon concentration; Based on primary suspicious index screening, intermediate spatiotemporal delay verification and advanced tensor decomposition, the diagnosis result of piston ring sticking is output; The coupling coefficient is updated online through a recurrent neural network to optimize the cost-sensitive loss function of the fault confusion matrix.

4. A diesel generator set shutdown detection and control system according to claim 3, characterized in that: The method of performing wavelet packet decomposition on the mechanical vibration signal to extract high-frequency sensitive frequency band energy and calculate the cylinder wall temperature field gradient tensor and hot spot diffusion index comprises the following steps: The mechanical vibration signal is decomposed by 8-layer wavelet packet to extract the energy characteristics of the piston ring sticking sensitive frequency band: ,in, For the wavelet system of the mth layer and the nth node, the sensitive frequency band corresponds to m=7, n=32-47, and the piston ring sticking sensitive frequency band is 2.5-4.2kH; Calculate the Frobenius norm of the cylinder wall temperature gradient tensor and generate the hot spot diffusion index ,in, is the heat conduction attenuation coefficient, and N is the number of effective grids.

5. A diesel generator set shutdown detection and control system according to claim 4, characterized in that: The nonlinear state space equation includes a mechanical vibration subsystem and a thermodynamic subsystem. The mechanical vibration subsystem includes ,in, is the mechanical vibration state variable, is the coupling term of thermal field and chemical field to mechanical vibration. The thermodynamic subsystem includes ,in, is the thermal diffusion coefficient, is the influence coefficient of blowby gas flow rate on temperature, is the influence coefficient of vibration energy on temperature. The cross-domain coupling coefficient identification uses Kalman filtering to estimate the coupling matrix online. ,in, and Represent the state variables of thermal field and chemical field respectively.

6. A diesel generator set shutdown detection and control system according to claim 5, characterized in that: The method of calculating the mutual information entropy of the vibration energy and the temperature gradient and constructing the dynamic transfer function of the blowby gas flow rate and the unburned hydrocarbon concentration includes the following steps: Calculate the mutual information entropy of the vibration energy and the temperature gradient, the mutual information entropy of the temperature gradient is the mutual information entropy of the hot spot diffusion index, ,like , it indicates that the mechanical-thermal coupling is abnormal; Constructing the dynamic transfer function of blowby gas flow rate and unburned HC concentration , a significant resonance peak appears in the 0.5-2Hz frequency band , indicating that there is a combustion abnormality.

7. A diesel generator set shutdown detection and control system according to claim 6, characterized in that: The method of outputting the diagnosis result of piston ring sticking based on primary suspicious index screening, intermediate spatiotemporal delay verification and advanced tensor decomposition includes the following steps: Primary screening calculation of piston ring sticking suspicion index ,in, is the weight coefficient, and is the reference value under normal working conditions. , then it is suspected that there is a piston ring sticking fault; Intermediate verification of the propagation delay of vibration energy and hot spot diffusion index ,like , eliminating transient interference and confirming it is a continuous fault; Advanced diagnostics builds a 4D fault feature tensor , using Tucker decomposition to extract core failure modes , Fault contribution ,like And other modes , it is diagnosed as a piston ring sticking fault.

8. A diesel generator set shutdown detection and control system according to claim 7, characterized in that: The method of updating the coupling coefficient online through a recursive neural network and optimizing the cost-sensitive loss function of the fault confusion matrix includes the following steps: Updating coupling coefficients using recurrent neural networks ,in, and It is the state change of mechanical vibration and thermal field; Define cost-sensitive loss function , give priority to reducing the false negative rate and complete the fault confusion matrix optimization.

Images